← Back to use cases
DHL targets inventory counts with machine learning
Added 2026-10-17 · Reviewed 2026-10-17 · DHL Supply Chain
Read the source ↓What the company did
DHL Supply Chain’s AI chief Christophe Theys told CIOnews that a machine learning model predicts where physical inventory may differ from system records. The predictions help target warehouse cycle counts, directing verification toward suspected discrepancies.
How it went
The October 16 interview describes an operating use, but gives no count-specific accuracy, labor saving or site coverage. It does not establish when the model went live. Company-wide AI and robot deployment totals do not measure this task’s performance.
Worth knowing
- The report does not specify model inputs, risk thresholds or how inventory adjustments are approved.
- Practical takeaway: Compare predicted risk with completed SKU-location counts, including unit-of-measure and recent-movement checks. Sample low-risk locations too, so discrepancies missed by the model remain visible before adjusting inventory.
Sources & provenance
DHL Supply Chain Scales Digital And Physical AI Through One Innovation Funnel ↗
CIOnews · Published 2026-10-16 · Reviewed 2026-10-17
“We call it ‘machine learning cycle count’”
Supporting evidence and review
Human review required
- Evaluate predicted risks against completed SKU-location counts. Keep some random checks to measure misses outside the targeted group.
Evidence & limitations
- No task-specific accuracy, count-time saving or deployment date disclosed.
- Firsthand company account, not an independent model evaluation.